Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y

www.eje-online.org © 2016 European Society of EndocrinologyPrinted in Great Britain

Published by Bioscientifica Ltd.DOI: 10.1530/EJE-16-0527www.eje-online.org © 2016 European Society of Endocrinology

175:6 595–603P J Donovan and others Cost–utility analysis of Graves’ disease therapies

European Journal of Endocrinology (2016) 175, 595–603

175:6

10.1530/EJE-16-0527

Cost–utility analysis comparing radioactive iodine, anti-thyroid drugs and total thyroidectomy for primary treatment of Graves’ diseasePeter J Donovan1,2, Donald S A McLeod2,3,4, Richard Little5 and Louisa Gordon4,6

1Department of Clinical Pharmacology, Royal Brisbane and Women’s Hospital, Herston, Queensland, Australia, 2School of Medicine and Biosciences, University of Queensland, Herston, Queensland, Australia, 3Department of Endocrinology and Diabetes, Royal Brisbane and Women’s Hospital, Herston, Queensland, Australia, 4Population Health Department, QIMR Berghofer Medical Research Institute, Herston, Queensland, Australia, 5Consultant Health Economist, Cambridge, UK, and 6Centre for Applied Health Economics, Logan Campus, Griffith University, Meadowbrook, Queensland, Australia

Clinical Study

Correspondence should be addressed to P J Donovan Email peter.donovan@health.qld.gov.au

Abstract

Objective: Little data is in existence about the most cost-effective primary treatment for Graves’ disease.

We performed a cost–utility analysis comparing radioactive iodine (RAI), anti-thyroid drugs (ATD) and total

thyroidectomy (TT) as first-line therapy for Graves’ disease in England and Australia.

Methods: We used a Markov model to compare lifetime costs and benefits (quality-adjusted life-years (QALYs)).

The model included efficacy, rates of relapse and major complications associated with each treatment, and alternative

second-line therapies. Model parameters were obtained from published literature. One-way sensitivity analyses were

conducted. Costs were presented in 2015£ or Australian Dollars (AUD).

Results: RAI was the least expensive therapy in both England (£5425; QALYs 34.73) and Australia (AUD5601; 30.97

QALYs). In base case results, in both countries, ATD was a cost-effective alternative to RAI (£16 866; 35.17 QALYs;

incremental cost-effectiveness ratio (ICER) £26 279 per QALY gained England; AUD8924; 31.37 QALYs; ICER AUD9687

per QALY gained Australia), while RAI dominated TT (£7115; QALYs 33.93 England; AUD15 668; 30.25 QALYs Australia).

In sensitivity analysis, base case results were stable to changes in most cost, transition probabilities and health-relative

quality-of-life (HRQoL) weights; however, in England, the results were sensitive to changes in the HRQoL weights of

hypothyroidism and euthyroidism on ATD.

Conclusions: In this analysis, RAI is the least expensive choice for first-line treatment strategy for Graves’ disease.

In England and Australia, ATD is likely to be a cost-effective alternative, while TT is unlikely to be cost-effective.

Further research into HRQoL in Graves’ disease could improve the quality of future studies.

Introduction

Graves’ disease is the most common cause of hyperthyroidism with an incidence of 0.8 cases per 1000 women annually in England (1). The three standard primary treatments have different profiles of potential benefits and harms for patients: radioactive iodine

(RAI) is associated with lower rates of relapse than anti-thyroid drugs (ATDs) but more potential to cause Graves’ ophthalmopathy (GO) (2, 3); ATDs can lead to long-term remission (with no deficits in quality of life and no long-term costs), but has high rates of relapse and is

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 596Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

associated with potentially catastrophic side effects (e.g. agranulocytosis); surgery (total thyroidectomy – TT) has the highest cure rates, but largest upfront costs and more potential long-term side effects (hypoparathyroidism, recurrent laryngeal nerve palsy and scar) (4). Therefore, to properly assess the cost-effectiveness of the primary therapies for Graves’ disease, modelling lifelong costs and effectiveness (measured by quality of life in quality-adjusted life-years – QALYs) is necessary. The few published cost-effectiveness analyses examining Graves’ disease management have important limitations including only short-term assessment of costs and benefits and/or assessment treatment options not consistent with contemporary management (i.e. subtotal thyroidectomy) (5, 6, 7, 8).

The primary aim of this study was to perform a cost–utility analysis comparing RAI, ATD and TT from the perspective of the government contribution to the healthcare sector, in both England and Australia.

Methods

Model structure

We conducted a cost–utility analysis using TreeAge Pro 2015 R2 (TreeAge Software, Williamstown, MA, USA). One model with the same structure was constructed for England and Australia, with the only differences being the inputs of costs, life-expectancy data and discount rates. Figure 1 shows a simplified version of the model. We used a Markov cohort, which is cyclical and tracks key clinical options and outcomes of persons with Graves’ disease following each of the three treatments.

Given peak age of onset of Graves’ disease is described as 40–60 years (9), and to capture the impact of it on younger patients, a 40-year-old female was selected as the base case patient. Life-expectancy data was obtained from recognised country-specific sources (10, 11). The Markov cycle length was three months, given that all transient or short-term states (e.g. transient hypoparathyroidism and symptomatic hyperthyroidism) should be near or fully resolved within this time frame. The maximum time horizon of the model was until age 100 years. Discount rates for costs and benefits beyond the first year were 3.5% per year in England and 5% in Australia, according to country-specific guidelines (12, 13). Carbimazole was used as the ATD of choice. In the base case analysis, for model simplicity all of the ATD cohort that relapsed after initial remission were treated with long-term ATD, rather than

definitive therapy (e.g. RAI or TT); the effect of differing proportions of RAI and TT as second-line therapy was assessed in sensitivity analysis in patients with relapse after initial ATD therapy. If RAI did not produce remission, retreatment with RAI (up to three doses) was included in the model, consistent with American Thyroid Association guidelines (8).

Rates of minor complications that are short-lived, have little effect on quality of life or lead to a change in treatment (e.g. minor rash or elevated liver enzymes from ATD), are rare (e.g. fulminant liver failure from propylthiouracil) or for which evidence to support causation is limited (e.g. RAI causing secondary cancers) were not included (3, 9). Although GO can occur at any time, only the excess risk of GO associated with RAI was included in the model (2). Additionally, although many women with Graves’ disease are of childbearing age, this aspect was excluded from this model, as, in patients where pregnancy is desired in the short-term, RAI is contraindicated due to potential teratogenicity (8).

Figure 1

Simplified version of the Markov model. ATD, antithyroid

drugs; GO, Graves’ ophthalmopathy; RAI, radioactive iodine.

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 597Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

Clinical estimates

We performed a literature review using PubMed and EMBASE to identify rates of efficacy, relapse, complications and HRQoL values, associated with each treatment option, with various combinations of the following search terms (PubMed terms only shown): hyperthyroid*, Graves, thyrotoxicosis, ‘Graves Disease’[Mesh], ‘Graves Disease/therapy’[MeSH], anti-thyroid drug, carbimazole, propylthiouracil, methimazole, (‘Antithyroid Agents’[Pharmacological Action], recurrence, radioiodine OR radioactive*, ‘Iodine Isotopes/therapeutic use’[MeSH], ‘Iodine Isotopes/therapy’[MeSH], complication, thyroidectomy[MeSH Terms], ‘cost utility’, ‘cost-utility’, QALY, EQ-5D, ‘quality adjusted life year’, ‘quality-adjusted life year’, ‘Quality of Life’[MeSH], ‘cost effective*’, ‘cost-effective*’, ‘cost utility’, ‘cost-utility’, ‘Cost-Benefit Analysis’[MeSH] and ‘economic evaluation’. Additional publications were obtained by targeted searches of the references of identified studies. Where more than one

potential publication was identified, meta-analyses and randomised trials were preferred (2, 3, 14). If only cohort studies were identified, data from studies with larger populations and longer durations of follow-up were included (15).

Table 1 shows a summary of the transition probabilities included in the model and their sensitivity analysis ranges. Transition probabilities are the probability of a patient transitioning from one Markov state to another, during a single Markov cycle (e.g. the probability that a patient transitions from active hyperthyroidism to hypothyroidism with TT, without having suffered any other complication of surgery, is 62.2% (Table 1)).

Costs

Unit costs were identified using recognised sources and presented in 2015 values (Pounds (Sterling); £ in England

Table 1 Transition probabilities and health-related quality-of-life weights.

Parameter valueSensitivity

analysis range Reference

Transition probabilities ATD Failure of ATD 5% over 1.5 years 0–20% Assumption Agranulocytosis 0.35% over 13 years 0.29–0.42% (15) Hypothyroidism 2.9% over 10.2 years 0–8.6% (25) Relapse post remission with ATD 52.8% over 3.73 years

(reverts to zero after 5 years)49.0–56.6% (3)

Radioactive Iodine Hypothyroid post-RAI 72.3% over 10 years 68.7–75.9% (26) Persistent Graves’ disease post first dose RAI 14.4% over 0.25 years 11.6–17.3% (26) Hypothyroid post second dose 77.5% over 0.25 years 73.2–81.7% (27) Hypothyroid post third dose 100% over 0.25 years – Assumption (27) Symptomatic GO Rate of 5.8% over 1.21 years

(reverts to zero after 15 months)2.5–9.1% (2)

Resolution of GO symptoms 99% resolution in 3 years 50–99.99% (14) Total thyroidectomy Hypothyroidism, no complications 62.2% over 0.25 years 57.1–67.4% (4) Hypoparathyroidism 33.0% over 0.25 years (resolves in 92%) 28.0–38.0% (4) RLN palsy 4.7% over 0.25 years (resolves in 69%) 2.4–6.9% (4) Health-related quality–of-life weights Remission 1.00 0.98–1.0 (24, 28) Euthyroidism while on ATD 0.98 0.96–1.0 (24, 28) Hypothyroidism after RAI (treated) 0.97 0.945–0.995 Expert opinion

using Delphi methodology

Hypothyroidism after TT (treated) 0.95 0.92–0.96 Expert opinion using Delphi methodology

Graves’ ophthalmopathy 0.88 0.86–0.90 (7, 28) Hypoparathyroidism 0.89 0.87–0.92 (29) Dysphonia from RLN palsy 0.89 0.87–0.92 (29) Thyrotoxicosis 0.81 0.78–0.82 (24, 28) Agranulocytosis 0.46 (for 7 days) 0.46–1.0 (30)

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 598Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

and 2015 Australian Dollars (AUD)) – see Table 2. Where unit costs were not readily available, estimates were obtained from published literature or by currency conversion, with costs from before 2015 adjusted to present values, where possible (6, 16, 17). The perspective taken for this analysis was of each governments’ contribution to healthcare.

Long-term costs of medications, medical practitioner visits and pathology associated with all treatments and their complications were included in the model, with the proposed follow-up visit schedule similar to previously published CEA and recognised treatment guidelines – see Table 3 (5, 8).

Health-related quality-of-life estimates

Effectiveness was evaluated by using HRQoL estimates (health utilities) from the published literature to generate QALYs obtained – Table 1. The preferred methods for obtaining HRQoL estimates to be included in the model were (in descending order of preference, consistent with country-specific guidelines (12, 13)):

• Preference-based methods with direct elicitation of HRQoL weights from prospective studies (e.g. time trade-off, standard gamble, multi-attribute utility indices with preference-based methodology (e.g. Euro-Quality of Life – 5 Dimensions (EQ-5D)))

• Studies using the SF-36 (Short Form 36) questionnaire and conversion of these scores to EQ-5D weights using a published, validated algorithm (24)

• HRQoL weights based on expert judgement, weights from previously published cost-effectiveness analysis or generated as part of this study (using Delphi methodology, seven specialist endocrinologists came to consensus values, after taking into consideration the other HRQoL weights used in the model).

Incremental cost-effectiveness ratio

The model aggregates the costs and patient outcomes using an expected values analysis, calculating the incremental cost-effectiveness ratios (ICER) using the following formula (ICER of ATD over RAI as an example):

ICER =-Lifetime cost of ATD Lifetime cost of RAI

Total QALYs AATD Total QALYs RAI-

In the main model, the ICER is calculated using the best available estimates, called the ‘base case’, while uncertainty and variation in these estimates are tested in sensitivity analyses (see below). The results were assessed for dominance (a dominant option is both less costly and more effective than another option) and extended dominance (where a treatment option is more costly and

Table 2 Unit costs in England and Australia.

England (2015 £) Australia (2015 AUD) References

Interventions and medicationsAdministration of RAI for treatment of Graves’ (including I-131,

per dose thyroid uptake scan, physician visits and ATD)556.85 341.15 (6, 17)

Thyroxine 4.04 (28 tablets) 24.02 (200 tablets) (18, 19)Same day admission for treatment of Graves’ ophthalmopathy

with intravenous methylprednisolone461 999 (17, 20)

ATD (carbimazole 5 mg) 76.49 (100 tablets) 31.38 (200 tablets) (18, 19)Admission with agranulocytosis 959 5076 (20, 21)Total thyroidectomy 2345 8344 (20, 21)Calcium carbonate (1.5 g tablets) 8.7 (100 tablets) 14.65 (120 tablets) (18, 19)Calcitriol (0.25 μg tablets) 25.76 (100 tablets) 30.62 (100 tablets) (18, 19)Total thyroidectomy with complications (i.e. RLN palsy or

hypoparathyroidism)2794 15 355 (20, 21)

Pathology and other investigationsThyroid function tests (TFTs) 12.93 34.80 (6, 16)Calcium studies (ionised) 4.47 9.70 (17, 22)Electrolytes and renal function tests 8.16 17.70 (17, 22)Thyroid uptake scan 243.51 175.40 (6, 16)

Medical attendancesSpecialist physician – initial (subsequent) 187.00 (93.00) 150.90 (75.50) (21, 22)Specialist surgeon – initial (subsequent) 140 (81.00) 85.55 (43.00) (21, 22)Ophthalmologist – initial (subsequent) 104 (59.00) 85.55 (43.00) (21, 22)General practitioner 45.00 37.05 (21, 22)

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 599Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

less effective than a combination of two other options). The National Institute of Health and Care Excellence (NICE) in England suggest a cost-effectiveness threshold of £20 000–30 000 per QALY gained (12). We chose a pre-specified threshold £30 000 per QALY gained in this analysis for England (and AUD 50 000), consistent with this guidance (12).

Sensitivity analyses

We performed one-way sensitivity analyses, where the value of a single parameter is changed across a range of values (the sensitivity analysis range) with different ICER values calculated. Comparing these values to the base case ICER can be used to assess how stable the results are to these changes, particularly in reference to the pre-specified thresholds for cost-effectiveness. Sensitivity analysis ranges of transition probabilities were based on 95% confidence intervals from published literature

(where available). Cost estimates vary from 50 to 150% depending on base case values. As no published data was available to assist with choice of sensitivity analysis ranges, the chosen ranges were arbitrary but were considered to be plausible. Age at entry to the cohort ranging from 20 to 60 years was also assessed.

Results

Base case

RAI was the least expensive therapy in both England (£5425; QALYs 34.73) and Australia (AUD5601, 30.97 QALYs). In both countries, ATD was a cost-effective alternative to RAI (£16 866, 35.17 QALYs, ICER £26 279 per QALY gained England; AUD8924; 31.37 QALYs; ICER AUD9687 per QALY gained Australia), while RAI dominated TT (£7115; QALYs 33.93 England; AUD15 668; 30.25 QALYs Australia).

Table 3 Follow-up schedule and healthcare utilisation.

State Medical practitioner visits Pathology per visit Medications

Hypothyroidism after any treatment

Every 6 months for 12 months (general practitioner)

TFTs Thyroxine 150 μg/day

GO Review with specialist ophthalmologist (once every 3 months while active)

6× weekly intravenous infusions of methylprednisolone (23) (day admission to hospital)

Initiation of ATD (first-line therapy for 18 months)

Every 6 weeks for 6 months (physician) Thyroid function tests (TFTs)

Carbimazole – 6 per day for 6 weeks, then 3 per day for 6 weeks, then 2 per day for 3 months, then 1 per day thereafter

Then every 3 months for 12 months TFTsContinue lifelong ATD Initially as per initiation of ATD

(for first 18 months, then every 6 months (physician))

TFTs Carbimazole – as above

Then every 12 months (general practitioner)

TFTs

Post TT (no other complications)

Initial 1 visit 2 months post-operatively (surgeon)

Thyroxine 150 μg/day

Then every 6 months for 12 months (general practitioner)

TFTs

Then every 12 months (general practitioner)

TFTs

Hypoparathyroidism (permanent) after TT

Initial 1 visit 2 months post-operatively (surgeon)

Thyroxine 150 μg/day, one calcium carbonate (600 mg) twice daily and one calcitriol twice daily

Then every 6 weeks for 6 months (physician)

TFTs, calcium studies

Then every 3 months for 12 months (physician)

TFTs, calcium studies

Then every 6 months (physician) TFTs, calcium studiesRLN palsy after TT Beyond first Markov cycle, no medical

costs in addition to post-TT hypothyroidism are likely to be expended even if RLN palsy remains permanent

Thyroxine 150 μg/day

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 600Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

Sensitivity analysis

Table 4 outlines the results of selected one-way sensitivity analysis. ATD was a cost-effective alternative to RAI in most sensitivity analyses (i.e. the calculated ICER ranges remained below the thresholds of cost-effectiveness – £30 000 in England, AUD50 000 in Australia), with the exceptions of HRQoL weights attached to hypothyroidism post-RAI, remission post-ATD and euthyroidism on ATD (where the calculated ICER ranges extended beyond these thresholds). ATD became more cost-effective (ICER became lower) as the proportion of the cohort that received second-line RAI increased following relapse after initially achieving remission with ATD. RAI was dominant over TT

(i.e. RAI was less costly and more effective) in sensitivity analysis of all parameters assessed.

Discussion

In our base case analysis, in England and Australia, RAI was the least expensive option for the primary treatment of Graves’ disease, while ATD was a cost-effective alternative to it. TT was more expensive and less effective than RAI (i.e. TT was dominated by RAI). These results were stable to changes in many key parameters and structural uncertainty tested in sensitivity analyses. Although TT was dominated and ATD was cost-effective

Table 4 One-way sensitivity analysis results.

Parameter Range tested

ICER £/QALY (over RAI) ICER AUD/QALY (over RAI)

ATD TT ATD TT

HRQoL weightHypothyroidism post-RAI 0.945–0.995 13 314-RAI dominant

(above 0.986)RAI dominant 4974-RAI dominant

(above 0.986)RAI dominant

Hypothyroidism post-TT 0.92–0.96 25 499–25 690 RAI dominant 9409–10 629 RAI dominantHyperthyroidism 0.78–0.82 26 203–26 453 RAI dominant 9657–9756 RAI dominantRemission 0.98–1.0 26 279–34 783 RAI dominant 9687–13 057 RAI dominantEuthyroid 0.96–1.0 15 855–91 303 RAI dominant 5978–29 542 RAI dominantHypoparathyroidism 0.87–0.92 26 237–26 317 RAI dominant 9679–9701 RAI dominantRecurrent laryngeal nerve palsy 0.87–0.92 26 255–26 296 RAI dominant 9678–9693 RAI dominantGraves’ ophthalmopathy 0.86–0.90 26 255–26 296 RAI dominant 9681–9692 RAI dominantAgranulocytosis 0.46–1.0 26 279–26 279 RAI dominant 9687–9687 RAI dominant

Transition probabilitiesFailure rate of primary ATD therapy 0–20% 24 730–33 583 RAI dominant 8460–15 444 RAI dominantRelapse rate following remission

with ATD95% CI 25 465–27 069 RAI dominant 9313–10 051 RAI dominant

Hypothyroidism post-ATD 95% CI 25 357–26 762 RAI dominant 9354–9862 RAI dominantExcess risk of GO with RAI 95% CI 26 594–26 954 RAI dominant 9026–10 330 RAI dominantGO resolves in three years 50–99.9% 26 233–26 283 RAI dominant 9639–9692 RAI dominantRate of hypothyroidism post-RAI 95% CI 26 178–26 375 RAI dominant 9653–9720 RAI dominantFailure rate of first dose RAI 95% CI 26 052–26 583 RAI dominant 9540–9884 RAI dominantFailure rate of second dose RAI 95% CI 26 196–26 360 RAI dominant 9639–9734 RAI dominantRate of relapse after achieving

euthyroid state with RAI95% CI 25 266–27 496 RAI dominant 9189–10 286 RAI dominant

Rate of RLN palsy 95% CI 26 257–26 301 RAI dominant 9669–9705 RAI dominantRate of hypoparathyroidism 95% CI 26 262–26 295 RAI dominant 9653–9720 RAI dominantRate of agranulocytosis 95% CI 26 262–26 294 RAI dominant 9666–9705 RAI dominant

CostsTotal cost of RAI 50–150% 24 352–28 205 RAI dominant 8263–11 110 RAI dominantCost of uncomplicated TT 50–150% 26 225–26 332 RAI dominant 9506–9868 RAI dominantCost of TT resulting in RLN palsy or

hypoparathyroidism same as uncomplicated TT

– 26 278 RAI dominant 9504 RAI dominant

Treatment cost of agranulocytosis 0 to base case 26 278–26 279 RAI dominant 9683–9689 RAI dominantCarbimazole cost 50–150% 25 581–26 976 RAI dominant 8230–11 143 RAI dominantTotal cost of treatment of GO

with methylprednisolone50–150% 26 355–26 202 RAI dominant 9156–10 218 RAI dominant

OtherAge, at entry to model 20–60 21 690–29 055 RAI dominant 7933–10 736 RAI dominantProportion having second-line RAI

(remainder ATD long-term)0–100% 7319–26 279 RAI dominant 4928–9687 RAI dominant

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 601Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

in both countries, the ICERs were different, with the cost per QALY gained being much less in Australia than England. These differences, as would be expected, are driven largely by differences in unit costs, because, other than life expectancy, discount rates and cost differences, the structure of the two models are identical. The main differences appear to be the cost of carbimazole (with the cost in England being about 10-fold higher than Australia, after adjustment for purchasing power parity), the cost of specialist endocrinology follow-up (2.5-fold higher) and TT costs (0.6-fold lower). In both models, all of these unit costs were obtained from reliable sources and likely reflect the true cost to the respective governments, and hence, are likely to represent true differences in the cost-effectiveness in these two countries.

This study is the first, using a lifetime horizon, to assess the cost-effectiveness of the three first-line therapies in Graves’ disease in England and Australia. Our results are consistent with a single centre England study that assessed cost per cure from hyperthyroidism (not just Graves’ disease), which captured all medical costs for two years post diagnosis (6). That study demonstrated that RAI was substantially cheaper over a short-term horizon than ATD or TT per cure; however, it did not assess long-term costs or quality of life. Our results are not consistent with a 2012 cost-effectiveness analysis from the United States that suggested surgery (subtotal thyroidectomy) was more effective and minimally more costly that RAI for Graves’ disease (7). We did not consider subtotal thyroidectomy as a treatment choice, given it is not a recommended therapy for Graves’ disease due to high rates of relapse (8). Other differences in our study were the inclusion of long-term costs, more recent data sources for transition probabilities and HRQoL weights, and the costs of medical services identified in England and Australia were much less than those in the previous study (7).

Although the results of this study were stable to variation in most key parameters, results were sensitive to changes in some HRQoL weights, particularly the weights attached to hypothyroidism post-RAI, remission following ATD and euthyroidism while on ATD. This sensitivity is potentially important for a number of reasons. First, in both countries, ATD ranged from being cost-effective compared with RAI, to being dominated by it across a relatively modest sensitivity analysis range. Similarly, in England, but not Australia, as the HRQoL of euthyroidism on ATD therapy decreased across its modest sensitivity analysis range, ATD became less cost-effective (with an ICER as high as £91 303 per QALY gained). Secondly, there are limited high-quality data (e.g. from prospective

studies using preference-based methods) to support many of the HRQoL weights used in this study and in particular, two of the weights were derived using expert opinion, albeit, the consensus opinion of seven specialist endocrinologists. Thus, further research into HRQoL estimates in thyroid disease (e.g. hypothyroidism, euthyroidism on ATD therapy) to obtain validated weights using more accepted methodologies (e.g. EQ-5D) could improve the accuracy of future economic models.

Our base case analysis included long-term ATD therapy as the treatment of choice for most patients with relapsed Graves’ disease. However, base case results were highly sensitive to the proportion of RAI therapy given as second line, as, with increasing proportions, ATD became highly cost-effective in both England and Australia (with ICERs down to £7319 and AUD4928). Therefore, given the uncertainty about long-term quality of life, if ATD is chosen as first-line therapy, second-line RAI (rather than long-term ATD therapy or TT) would appear to be the most cost cost-effective therapy in the event of relapsed Graves’ disease.

There are a number of other limitations to this study. First, the acquisition of unit cost data, particularly in England, was problematic, as lists of unit costs, particularly for pathology and radiology services, are not readily available. However, findings were insensitive to a wide range of changes in unit costs and therefore appear to be stable to these uncertainties. Secondly, although a thorough sensitivity analysis was performed, its extent was limited by the available data. For example, the sensitivity analysis ranges for HRQoL and cost data were arbitrary and, although we believe them to be plausible, this is open to interpretation. Thirdly, there are situations where a particular therapy may not be a valid first-line choice, which are not accounted for in this model. For example, the use of RAI in women of childbearing potential, particularly those that are interested in pregnancy in the short-term is contraindicated due to possible teratogenicity, while thyroidectomy might be favoured in large goitres or if there are concerns about thyroid cancer (8). In addition, despite the inclusion of GO in the model, many clinicians might be reluctant to give RAI if a patient had severe, active GO, preferring an alternative therapy that does not have the potential to cause worsening symptoms (8). Fourthly, this study was performed from the perspective of the government contribution to the healthcare sectors in each country and thus ignores any costs borne by patients (e.g. out-of-pocket medication costs), their preferred choice of therapy and

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 602Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

any anxiety that may be experienced (e.g. as a result of possible future cancer risk from RAI). Further, the results of our study may not be readily generalizable to other countries, given that local practice and the costs of medical services are likely to differ.

Finally, the choice of cost-effectiveness threshold (i.e. how much one is willing to pay for one extra QALY) is potentially important, particularly in the English analysis. NICE guidance recommends a cost-effectiveness threshold of £20 000 to £30 000 per QALY gained (12). If the threshold was rigidly set at £20 000 per QALY gained, ATD may not be cost-effective in England, because in the base case and in most one-way sensitivity analyses, the ICER estimates sit between £20 000 and £30 000 per QALY gained. NICE guidance suggests that a threshold of £30 000 per QALY gained may be used where there is some uncertainty around the true ICER and HRQoL capture. As there is uncertainty in both these in our study, we believe that our pre-specified threshold of £30 000 per QALY gained is reasonable.

In conclusion, in this cost–utility analysis, RAI is the least costly first-line treatment of Graves’ disease in both England and Australia, while ATD, but not TT, may be a cost-effective alternative. These results are robust to substantial sensitivity analysis of cost and transition probabilities. However, the results are potentially sensitive to changes in some HRQoL weights, particularly hypothyroidism post definitive therapy and euthyroidism on ATD therapy. Where ATD is chosen as first-line, RAI as second-line therapy in the event of a relapsed Graves’ disease is likely to be more cost-effective than long-term ATD or TT. Further research into the HRQoL of many of the disease states associated with Graves’ disease and its treatment complications, and taking a wider (e.g. societal) perspective for analysis could add to the quality, robustness and comparability of future CEA in Graves’ disease.

Declaration of interestThe authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

FundingAn NHMRC Early Career Fellowship (APP1092153) supports Donald McLeod.

AcknowledgementsThe authors acknowledge Prof E Duncan; Associate Professor Michael d’Emden; and Drs S Lazarus, D Perry-Keene, C Baskerville and M Keogh for their assistance with proving health-related quality-of-life values used in this analysis.

References 1 Vanderpump MP, Tunbridge WM, French JM, Appleton D, Bates D,

Clark F, Grimley Evans J, Hasan DM, Rodgers H, Tunbridge F et al. The incidence of thyroid disorders in the community: a twenty-year follow-up of the Whickham Survey. Clinical Endocrinology 1995 43 55–68. (doi:10.1111/j.1365-2265.1995.tb01894.x)

2 Acharya SH, Avenell A, Philip S, Burr J, Bevan JS & Abraham P. Radioiodine therapy (RAI) for Graves’ disease (GD) and the effect on ophthalmopathy: a systematic review. Clinical Endocrinology 2008 69 943–950. (doi:10.1111/j.1365-2265.2008.03279.x)

3 Sundaresh V, Brito JP, Wang Z, Prokop LJ, Stan MN, Murad MH & Bahn RS. Comparative effectiveness of therapies for Graves’ hyperthyroidism: a systematic review and network meta-analysis. Journal of Clinical Endocrinology and Metabolism 2013 98 3671–3677. (doi:10.1210/jc.2013-1954)

4 Guo Z, Yu P, Liu Z, Si Y & Jin M. Total thyroidectomy vs bilateral subtotal thyroidectomy in patients with Graves’ diseases: a meta-analysis of randomized clinical trials. Clinical Endocrinology 2013 79 739–746. (doi:10.1111/cen.12209)

5 In H, Pearce EN, Wong AK, Burgess JF, McAneny DB & Rosen JE. Treatment options for Graves disease: a cost-effectiveness analysis. Journal of the American College of Surgeons 2009 209 170–179. e171–e172. (doi:10.1016/j.jamcollsurg.2009.03.025)

6 Patel NN, Abraham P, Buscombe J & Vanderpump MPJ. The cost effectiveness of treatment modalities for thyrotoxicosis in a U.K. center. Thyroid 2006 16 593–598. (doi:10.1089/thy.2006. 16.593)

7 Zanocco K, Heller M, Elaraj D & Sturgeon C. Is subtotal thyroidectomy a cost-effective treatment for Graves disease? A cost-effectiveness analysis of the medical and surgical treatment options. Surgery 2012 152 164–172. (doi:10.1016/j.surg.2012.02.020)

8 Bahn Chair RS, Burch HB, Cooper DS, Garber JR, Greenlee MC, Klein I, Laurberg P, McDougall IR, Montori VM, Rivkees SA et al. Hyperthyroidism and other causes of thyrotoxicosis: management guidelines of the American Thyroid Association and American Association of Clinical Endocrinologists. Thyroid 2011 21 593–646. (doi:10.1089/thy.2010.0417)

9 Brent GA. Clinical practice. Graves’ disease. New England Journal of Medicine 2008 358 2594–2605. (doi:10.1056/NEJMcp0801880)

10 United Kingdom, National Life Tables, 1980–82 to 2011–13. Newport, UK: Office of National Statistics, 2013.

11 3302.0.55.001 – Life Tables, States, Territories and Australia, 2011–2013. Canberra, Australia: Australian Bureau of Statistics, 2013.

12 Guide to the methods of technology appraisal 2013. London, UK: National Institute for Health and Care Excellence, 2013.

13 Guidelines for preparing submissions to the Pharmaceutical Benefits Advisory Committee. Canberra, Australia: Department of Health and Aging, 2015.

14 Jarhult J, Rudberg C, Larsson E, Selvander H, Sjovall K, Winsa B, Rastad J, Karlsson FA & Group TEOS. Graves’ disease with moderate-severe endocrine ophthalmopathy-long term results of a prospective, randomized study of total or subtotal thyroid resection. Thyroid 2005 15 1157–1164. (doi:10.1089/thy.2005.15.1157)

15 Tajiri J & Noguchi S. Antithyroid drug-induced agranulocytosis: special reference to normal white blood cell count agranulocytosis. Thyroid 2004 14 459–462. (doi:10.1089/105072504323150787)

16 Curtis L, Ed. Unit Costs of Health and Social Care 2013. University of Kent, Canterbury, UK: Personal Social Services Research Unit, 2013.

17 Purchasing Power Parity Dataset. Paris, France: Organisation of Economic Co-operation and Development (OECD), 2015.

18 Committee JF. British National Formulary. London, UK: British Medical Association and Royal Pharmaceutical Society of Great Britain, 2015.

19 Pharmacuetical Benefits Schedule. Canberra, Australia: Department of Health and Aging, 2015.

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access

Euro

pea

n J

ou

rnal

of

End

ocr

ino

log

y175:6 603Clinical Study P J Donovan and others Cost–utility analysis of Graves’

disease therapies

www.eje-online.org

20 Australian refined diagnosis-related groups (AR-DRG) – Cost Weights for AR-DRG Version 6.0x (2011–2012) current version. Bruce, Australia: Australian Institute of Health and Welfare, 2011–2012.

21 National tariff payment system 2014/15. London, UK: National Health Service, 2015.

22 Medical Benefits Schedule. Department of Health and Aging, 2015. 23 Bartalena L, Baldeschi L, Dickinson AJ, Eckstein A, Kendall-Taylor P,

Marcocci C, Mourits MP, Perros P, Boboridis K, Boschi A et al. Consensus statement of the European group on Graves’ orbitopathy (EUGOGO) on management of Graves’ orbitopathy. Thyroid 2008 18 333–346. (doi:10.1089/thy.2007.0315)

24 Ara R & Brazier J. Deriving an algorithm to convert the eight mean SF-36 dimension scores into a mean EQ-5D preference-based score from published studies (where patient level data are not available). Value Health 2008 11 1131–1143. (doi:10.1111/j.1524-4733.2008.00352.x)

25 Azizi F, Ataie L, Hedayati M, Mehrabi Y & Sheikholeslami F. Effect of long-term continuous methimazole treatment of hyperthyroidism: comparison with radioiodine. European Journal of Endocrinology 2005 152 695–701. (doi:10.1530/eje.1.01904)

26 Hernandez-Jimenez S, Pachon-Burgos A, Aguilar-Salinas CA, Andrade V, Reynoso R, Rios A, Reza-Albarran AA, Mehta R,

Gonzalez-Trevino O, Gomez-Perez FJ et al. Radioiodine treatment in autoimmune hyperthyroidism: analysis of outcomes in relation to dosage. Archives of Medical Research 2007 38 185–189. (doi:10.1016/j.arcmed.2006.09.007)

27 Metso S, Jaatinen P, Huhtala H, Luukkaala T, Oksala H & Salmi J. Long-term follow-up study of radioiodine treatment of hyperthyroidism. Clinical Endocrinology 2004 61 641–648. (doi:10.1111/j.1365-2265.2004.02152.x)

28 Elberling TV, Rasmussen AK, Feldt-Rasmussen U, Hording M, Perrild H & Waldemar G. Impaired health-related quality of life in Graves’ disease. A prospective study. European Journal of Endocrinology 2004 151 549–555. (doi:10.1530/eje.0.1510549)

29 Sejean K, Calmus S, Durand-Zaleski I, Bonnichon P, Thomopoulos P, Cormier C, Legmann P, Richard B, Bertagna XY & Vidal-Trecan GM. Surgery versus medical follow-up in patients with asymptomatic primary hyperparathyroidism: a decision analysis. European Journal of Endocrinology 2005 153 915–927. (doi:10.1530/eje.1.02029)

30 Perlis RH, Ganz DA, Avorn J, Schneeweiss S, Glynn RJ, Smoller JW & Wang PS. Pharmacogenetic testing in the clinical management of schizophrenia: a decision-analytic model. Journal of Clinical Psychopharmacology 2005 25 427–434. (doi:10.1097/01.jcp.0000177553.59455.24)

Received 21 June 2016Revised version received 8 August 2016Accepted 15 September 2016

Downloaded from Bioscientifica.com at 12/09/2020 12:11:36PMvia free access